MatLab Training Course Notes
Session 2

Instructor: Bill Miller
Dynamic Neuroimaging Laboratory
415.502.3726
http://dnl.ucsf.edu

Topics for Session 2

1. Suggested reading: Introductory MatLab books
2. Review of variables and syntax
3. Basic file loading and saving

1. Suggested reading

The help and reference information that comes with MatLab is essentially a list and description of functions and functionality. All their documentation unfortunately doesn't really teach you about how to use MatLab. If you want something more like a guide to how to use MatLab (more than my online notes), you might try one or more the books on this list. In addition, since MatLab includes its own versions of some useful commands from the C programming language, you may find a C-language reference book useful (also on the above linked page).

• "function syntax. MatLab does things with your data using "functions", which may be built-in or written by you (sometimes these are called commands since you may type them at the command line. Functions take "arguments" (or parameters) following the function name. If the argument(s) are strings (text), you don't need to put them in parentheses, but if they are numbers or names of variables, you do need to put them all inside a pair of parentheses. If inside parentheses, arguments are separated by commas.

     
        Examples
          cd mydirectory	no parens, string argument
          plot(a)	parens, non-string argument
          plot(x,y)	two arguments separated by a comma
  

• whos lists current variables in the workspace, with their size and type

• Arithmetic, algebra and variables: You can type math at the command line just as you might write it on paper:

          1+2		adds 1+2 and displays the result (in a default variable called "ans")
          a=1+2	creates the variable "a", and assigns it the value 3
          b=a+10	creates "b", and assigns it the value 13 (assuming a=3)
  

• [] defines vectors and arrays

       a=[2 4 6]	defines "a" as a vector with three values
       a=[]		"a" is variable with "null" (or empty set) as its value
   

• () parentheses are used for
  - specifying function arguments, and
  - doing "indexing" (selecting an element or elements from a vector or array)

        a=rand(1,10);		create a 1-row, 10-column array of random numbers (a built-in function)
        b=a(3)		b contains the third element of a
   

• ; The semicolon has two uses:
  - define a new row in an array
  - turn off display of result

      a = [1 2 3;4 5 6];        a is a two-row, three-column matrix. MatLab is silent when you type this.
  

• : The colon has two uses:
  - to create a range of values for a vector, and
  - to refer to all rows or columns of an array

    a= [0:0.1:10]		a contains the values 0, 0.1, 0.2, ... 10
    b = [2 4 6;8 10 12];          if "b" is defined this way, then:
    b(:,1)			returns the values in all rows of first column of b
    b(1,2:end)		returns the values in columns 2-3 of first row, using "end"
      

  --> note from the above example that arrays are indexed as "variable_name(rows,cols)"

• ' The single quote (apostrophe) has two uses:
  - transpose a vector or array
  - create a string

      a=[1:4], b=a'		b is the transpose of a
      c='bill was here'	c is a string (type "whos" to see the type)
  

  --> note in the first example above, a comma (or semicolon) can separate two or more commands on the same line.

• find This is a new function we didn't talk about in Session 1: find extracts the indexes of elements of a vector/array according to logical criteria

      a=rand(1,10);   a is a row of random numbers
      b=find(a<.5)	b are indexes of values in "a" which are less than .5
      a(b)		obtain the values of those selected elements (indexes in b)
      c=a(b)              c is a new vector containing only those selected elements
  

  Here are the basic logical operators: == (equals), ~= (not equals), <, >, <=, >=
  Type "help ops" or "help ==" etc. for the whole list.

• text files, binary files, mat files
  • text files have ASCII codes - numbers and delimiters, maybe with a header.
  • binary files just have 1's and 0's - you have to know the format (how big a number is and how many)
  • mat files are matlab binary files: they store the data and the variable names.
• => test (how to tell the difference): can you read your file with a text editor like Notepad?

• load This function is useful for loading mat files and many text simple files

     load flujet	(A Matlab demo image; you don't need to use the ".mat" extension)
     whos		(look at all the variables just loaded from the one file)
     load temp1.txt  (creates the variable "temp1" from the file name
            	     and puts the data from the file in there
            	     - works with simple space delimited text files with no header)
  

  --> here is the temp1.txt file to try. (Right click on the link and save it to a file with that name in a directory on your MatLab path).

• save saves MatLab binary or text files. The format is:

        save [filename] [variable(s)]
  
     Example
     a=1;b=2;c=3;
     save myvars a b c	saves all three variables in a mat file 
                      	called "myvars.mat" (extension not required)
  
     clear              	clear all variables
     load myvars a b      load just "a" and "b" from myvars.mat
  

• loading with C-language equivalents MatLab includes a number of functions that are equivalent to those used in C for opening, reading and writing files. The main difference is that in MatLab, an entire file can be read or written in one step, while in C (and other programming languages), one reads or writes a file number by number or line by line. This is another example of how MatLab is "vectorized". Using these functions requires more steps than "load" and "save", but gives you more control.

  The most important file loading and saving functions are:

  • fopen, fclose these open and close files
  • fscanf, fprintf for reading and writing text files
  • fread, fwrite for reading and writing binary files

  For brevity, we will just look at examples of reading files. Type "help iofun" for details about these and many more file input/output (IO) functions.

      Examples
  
       Text file 
         (temp1.txt has two columns 
         of floating point numbers, tab delimited)
   
        fid=fopen('temp1.txt','rt');	open temp1.txt for "read text"
        a=fscanf(fid,'%f %f',[2,101])'	read all data into "a" (two rows by 
                                     	  101 columns, then transposed) 
        fclose(fid)                 	close file
        plot(a(:,1),a(:,2))            	plot col 2 (y) against col 1 (x) - sine wave!
  

  --> Note in the fscanf statement that we read in the values in the values assuming the data was 2 rows and 101 columns, and then transposed them, although the data was really saved as many rows and 2 columns. This is one of the relatively few non-intuitive instances of MatLab function design.

       Binary file 
         (temp1.bin was created with
           101x2 float values, same data as temp1.txt)
  
        fid=fopen('temp1.bin','rb');	open temp1.bin for "read binary"
        b=fread(fid,[101,2],'float');	read all data into "b"
        fclose(fid);                 	close file
        plot(b(:,1),b(:,2))               plot the sine wave (x vs. y)
  
       MatLab binary file
        load temp1.mat               	temp1.mat contains one variable
                                     	  "c" with the sine wave data (I've included the ".mat"
                                   	  extension to allow MatLab to distinguish it 
                                     	  from "temp1.txt" in this particular example)
        plot(c(:,1),c(:,2))
  

  In summary, the above examples show how the same data can be loaded into MatLab although stored in files in different formats. Once loaded into a workspace variable in MatLab, the data is treated the same way (i.e., you can basically forget about how you loaded the data).

  There are two other very useful functions for text file reading.

  • hdrload: If a text file has one or many initial "header" lines of text (the initial character of each line is not a number), this is a nice function found on the MathWorks website that separates the header and the data into separate variables called hdrload.m. (Right click on the link and save it to a file with that name in a directory on your MatLab path).

  • dlmread: If your text file has data delimited with commas (and NO HEADER), use this installed function.

  Finally, imread is a good basic function for loading pictures/images in many of the standard formats (jpeg, bmp, etc.).

  In the next session, we will look at more about plotting data, creating figures with several plots on them, and working with figure properties; and introduce the basics of programming and function creation in MatLab.